-
The launch of the AI RAN Alliance last week took a lot of people by surprise
-
Founding members include Nvidia and SoftBank but some AI and ML bigwigs are missing
-
Mavenir, which was not among the founding members, touts its open RAN Intelligent Controller (RIC) with AI and ML
Most of the artificial intelligence (AI) that wireless operators are doing has to do with the customer experience as opposed to the radio access network (RAN). That’s in part why Nvidia, SoftBank and a handful of other companies launched the AI-RAN Alliance at Mobile World Congress in Barcelona last week.
While Nvidia is the hottest ticket in AI chips today, it’s worth noting who’s not on the list, including CommScope, Intel, NEC, Fujitsu, Mavenir, Qualcomm and more operators.
The AI-RAN Alliance’s mission is to reduce power consumption, enhance mobile network efficiencies and unlock new revenue opportunities for telecom companies through AI, both in 5G and 6G. Founding members are Amazon Web Services (AWS), Arm, DeepSig, Ericsson, Microsoft, Nokia, Northeastern University, Nvidia, Samsung Electronics, SoftBank and T-Mobile USA.
“To me, the biggest omission is the lack of more mobile network operators,” said Recon Analytics analyst Daryl Schoolar. “The alliance launched with just SoftBank and T-Mobile.”
Other mobile operators are welcome to join, including AT&T and Verizon, according to Ryuji Wakikawa, head of SoftBank’s Research Institute of Advanced Technology. In an interview last week, he emphasized that it’s an open alliance and they want to develop an entire ecosystem. Others are expected to join their efforts.
However, it doesn’t sound as though Qualcomm, which is pushing AI in phones big time, will be one of them – at least, not immediately.
“Qualcomm believes AI will play a crucial role in 5G and 6G infrastructure. In fact, prior to MWC we announced Qualcomm Infrastructure Processors with the aim to redefine AI and RAN workload processing in vRAN servers,” said Gerardo Giaretta, VP of product management at Qualcomm Technologies, in a statement provided to Fierce. “While we are not members of this particular alliance, we share a common goal of transforming 5G and 6G infrastructure through AI, so we will monitor their progress and continue to work towards that goal.”
What about Mavenir and Intel?
The lack of a personal invitation to join the alliance was particularly un-nerving for open RAN pioneer Mavenir, in part because Nvidia invested in Mavenir in 2020, revealed Mavenir CEO Pardeep Kohli to Fierce at Mobile World Congress (MWC) last week.
To top things off, the AI-RAN Alliance’s display in the Arm booth was directly across from Mavenir’s booth. From any number of tables in Mavenir’s lobby, you could get a clear view of the AI RAN Alliance demo display – a kind of “in your face” taunt, if you will.
Asked about the alliance, Kohli acknowledged that Mavenir wasn’t consulted about the AI-RAN Alliance prior to MWC. Still, he said Mavenir will be joining the alliance, which he views as a positive step for the industry.
Interestingly, Kohli said Mavenir is already doing AI in the RAN. “We are doing it for enterprise applications today,” Kohli said. “We are also testing it for macro” deployment.
Similar to open RAN, bigger companies are getting on the bandwagon, and that’s good in the sense that they can explain it to everyone else. “Hopefully, some people will buy from them and some people will buy from us,” he said. “We will all be speaking the same language.”
Mavenir says its open RAN Intelligent Controller (RIC) is uniquely designed with deep knowledge of the RAN domain, AI and machine learning (ML). Bharti Airtel in India is the first operator to use Mavenir’s AI solution to improve network performance.
“What AI and ML can do that humans can’t do is they can mine all that complex telco data and they can start extracting patterns and insights,” said Brandon Larson, general manager of Mavenir’s Cloud, AI and IMS Business Strategy. Once more is understood about the network, “we can actually tune the performance better and start to make recommendations on things like power settings or handover thresholds and things like that," he said.
Larson likens it to Google Maps, where data on traffic patterns and algorithms are used to suggest the best route. “We’re effectively doing that with telco networks,” he said, using AI and ML to crunch data and provide recommendations to operations staff on how they should handle handovers or improve spectral efficiencies, for example.
Besides Airtel, Mavenir is talking to about a dozen other operators, including in North America, about incorporating AI/ML into their networks – basically, “everywhere you see open RAN.” But first, they want to “get everything done right” with one operator before expanding, he said.
Another player clearly missing from the alliance is Intel, which used space in its MWC booth to tout new AI solutions for the RAN that span network and edge AI. The vendor noted that artificial intelligence will play a “pivotal role” in helping operators optimize the performance of resources in the evolving vRAN environment.
Intel is making its vRAN AI Development Kit available to select partners, and it’s working with AT&T, Deutsche Telekom, SK Telecom and Vodafone to showcase the benefits AI can bring to the RAN.
AI-RAN: Not yet ready to run
AI built natively into RAN equipment is probably still years away, although there’s an opportunity to start from ground zero when 6G comes along.
Recon Analytics’ Schoolar noted the different ways of doing AI. For example, it’s plausible that AI will be used in Massive MIMO to make that more efficient.
The debate is sure to continue over whether something it “real AI” or “fake AI,” and/or merely someone throwing the term into their marketing literature to be considered part of the AI conversation.
One thing is for sure. AI is going to be everywhere, from the wireless network to devices.
“It’s going to be in all parts of the mobile ecosystem,” Schoolar said. “It is essentially like a seasoning that people are going to add to every dish. You’re going to see it in the customer experience, customer management. You’re going to see it in network operations. You’re going to see it embedded in the actual infrastructure solution as well, and you’re going to see it in devices too.”